The last generation of transceivers (i.e. communication equipments comprising a transmitter and a receiver) is capable of supporting multiple data rates either by changing the modulation format, or the coding rate or else the symbol rate of signals. This allows the transceiver to switch from one data rate to another one, not only between two signal transmissions or receptions but also during a signal transmission or reception.
As known by those skilled in the art, this last case requires the transceiver to adapt itself on the fly without losing any signal data, which means that the BER (Bit Error Rate) must be maintained below a given threshold. Such an adaptation requires an update of some of the digital signal processing (or DSP) software blocks into the receiver, which takes a transition time to reach a good convergence with a stable BER performance.
For instance, the change of symbol rate requires the update of DSP blocks such as the polarization demultiplexing and equalization (for instance employing the MMA (Multi-Modulus Algorithm)), the frame alignment and phase synchronization, and such a move takes a relatively long convergence time. MMA algorithms for instance implement stochastic gradient methods.
So, during a transition phase, i.e. before reaching convergence, the BER performance can be degraded and can fluctuate, but there is no known solution for overcoming these drawbacks.